Fruit classifier that can classify 120 different fruits and vegetables using fast.ai library.
The model used is based on resnet34 and after fine tuning, the model reached 99.4% accuracy on the validation dataset.
- fastai library
- Kaggle account
All instructions can be found in the Jupyter notebook. Use GPU hardware accelerator for shorter training times.
Data for training was taken from Kaggle: Fruits 360.
You can also use your own dataset. You can refer to the fast.ai documentation for details on how to load your own dataset.
As mentioned in the notebook, you can download and store the dataset on the runtime server or on Google Drive by mounting it.
Note: The Jupyter notebook was run using Google Colab. Therefore it is recommended, you use the same.